Since the institution has 4000 undergraduate students, we can take a sample of 20% of the population, that is, 800 students.

Moreover, the institution will have many courses and thus we already have natural groups that separate our given population. So the best method to use for sampling would be cluster sampling. We have natural groups that are separated by different courses and hence each course would be a representative of its sample. Applying cluster sampling we can choose the subgroups and can further sample out the students from each subgroup using simple random sampling. We can also use systematic sampling instead of simple random sampling that would sample the students according to some pre-defined rule.